Regulatory uncertainty Meta will not offer new multimodal AI models in the European Union

According to the financial news app Zhitong Finance, Meta Platforms (META.US), the parent company of Facebook, has stated that due to the unpredictable regulatory environment in Europe, it will not provide its multimodal artificial intelligence model Llama to EU customers. Last month, Meta announced a delay in the launch of its artificial intelligence assistant Meta AI in Europe as it deals with requests from the Irish privacy regulator regarding the use of Facebook and Instagram content as training data. Meta also faced complaints from the advocacy group NOYB in 11 European countries, accusing the company of planning to use personal data to train its AI models without user consent. Similarly, Apple (AAPL.US) indicated last month that it has postponed the release of several AI features in the EU due to the Digital Markets Act (DMA). The DMA is legislation that establishes objective criteria for identifying large online platforms as "gatekeepers" and ensuring their fair conduct online, leaving room for dispute. Meta plans to integrate the new multimodal model into various products, including smartphones and the company's Meta Ray-Ban smart glasses, which can infer video, audio, images, and text. Meta pointed out that European businesses and institutions are likely to fail as they cannot access the latest and best open models. The company added that European regulators have taken longer to determine legal requirements compared to other regions worldwide. Meta also stated that the General Data Protection Regulation (GDPR) in the EU and the UK are nearly identical, but the company continues to operate in the UK because UK authorities have provided clear feedback on identified issues. Therefore, according to sources, this is not a legal issue but rather a problem with the unpredictable application of laws in the EU. Meta also plans to soon launch a larger text-only version of the Llama 3 model. According to the company, this model will be introduced to EU customers and companies. Training AI models requires a large amount of data, and EU companies must comply with EU regulations on how to collect and process user personal data. In May, Meta announced its intention to use public posts from Facebook and Instagram users to train future models. The company stated that it had informed EU regulators of its AI plans months before the announcement and addressed their feedback, although feedback was minimal. In June of this year, the Irish Data Protection Commission (DPC) on behalf of European data protection authorities requested Meta to postpone using public content on Facebook and Instagram to train its large language model (LLM). Meta added that after the launch, EU regulators requested a pause without explicitly expressing concerns and posed over 270 detailed questions. The company noted that it will continue to engage with the Irish DPC, but the reality of the European regulatory environment is that it is unclear when training can begin However, the company pointed out that training on European Union data is crucial to ensure that its products accurately reflect the region's terms and culture. Meta stated that its models would not work as well for European citizens and businesses, as they would become more Americanized rather than globalized. Meta had previously stated that it is following the lead of other companies, such as Google under Alphabet and OpenAI supported by Microsoft, both of which have used data from Europeans to train artificial intelligence